Elucidating the vulnerability of grassland ecosystems and the spatial driving mechanisms under global change and human intervention is the scientific basis for understanding grassland ecosystem changes and effectively responding. In this study, we quantified the sensitivity, adaptability and vulnerability of grassland ecosystems across different eco-geographical zones of the Yellow River Basin based on vegetation net primary productivity (NPP), its inter-annual variability and trends. The spatial distribution patterns and driving factors of grassland ecosystem vulnerability in the Yellow River Basin were revealed.by combining geo-detector and geographically weighted regression models. The results show that: (1) The vulnerability of grassland ecosystem in the Yellow River basin is generally high, with the area of moderate and higher vulnerability accounting for 61%. In terms of spatial distribution, the vulnerability of grassland ecosystem shows a trend of increasing from southeast to northwest. (2) There are significant differences in the vulnerability of grassland ecosystem in different eco-geographical areas of the Yellow River basin. Hetao has the highest vulnerability with the desert steppe of the central plain of western Inner Mongolia (IID1), the central plain of eastern Inner Mongolia (IIC3) and the wide valley alpine meadow steppe of the southern Qinghai Plateau (HIC1). Deciduous broad-leaved forest in North China mountain (HIB1), loess beam and mountain grassland (IIIC1), cultivated vegetation in North China Plain (IIIB2), coniferous forest and grassland in Qilian Qingdong alpine basin (HIIC1). The vulnerability of Fen-Wei basin and southern Loess Plateau deciduous broad-leaved forest, cultivated vegetation area (IIIB4), middle Shandong low mountain hilly deciduous broad-leaved forest, cultivated vegetation area (IIIB1), North China mountain deciduous broad-leaved forest area (IIIB3) is the lowest. (3) The main driving factors of grassland ecosystem vulnerability in the Yellow River basin are NDVI and annual precipitation, and the interaction among all driving factors is significantly enhanced. (4) There is a predominantly negative correlation between average annual precipitation, NDVI, altitude, and grassland ecosystem vulnerability, and the impact of NDVI on grassland ecosystem vulnerability was spatially non-equilibrium. The study shows that the climatic conditions of the geographical climatic zone and the productivity characteristics of the ecological background directly determine the spatial distribution characteristics of grassland ecosystem vulnerability in the Yellow River basin.